What Dogs Know About Trails: Metaphor, Mechanisms, and the Phylogeny of Cross-Domain Extension

Lakoff and Johnson’s (1980) Conceptual Metaphor Theory and Boyd’s (1979) Homeostatic Property Cluster theory developed in parallel without mutual engagement, but both reject classical categories and invoke property clustering. This paper argues that HPC theory reframes CMT by supplying the stability mechanisms that Lakoff and Johnson’s account lacks. Metaphor, on this view, is cross-domain HPC extension on partial property overlap. The reframing recasts several standing disputes in metaphor theory, provides what CMT needs (a principled account of why some metaphors are productive and others superficial), and predicts the full range from literary metaphor through everyday to theory-constitutive, on a single dimension: the degree to which mechanisms in both domains sustain projectible inference. A case study of canine trail cognition illustrates the framework and generates a testable prediction: if trail is an HPC kind, dogs should recruit it for objects sharing only visual-geometric trail properties, even across functional domains.